CN104658254B - Motorcycle detection method for traffic videos - Google Patents
Motorcycle detection method for traffic videos Download PDFInfo
- Publication number
- CN104658254B CN104658254B CN201510102547.7A CN201510102547A CN104658254B CN 104658254 B CN104658254 B CN 104658254B CN 201510102547 A CN201510102547 A CN 201510102547A CN 104658254 B CN104658254 B CN 104658254B
- Authority
- CN
- China
- Prior art keywords
- motorcycle
- detection
- color
- confidence
- lbp
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Landscapes
- Image Analysis (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention relates to a motorcycle detection method for traffic videos. According to the method, a plurality of detection frames are acquired in a window screening manner. the method specifically comprises steps as follows: 1) extracting the HoG feature from each detection frame to acquire the HoG confidence; 2) extracting the LBP feature from each detection frame to acquire the LBP confidence; 3) extracting the color feature from each detection frame to acquire the color confidence; 4) obtaining motorcycle detection results according to the HoG confidence, the LBP confidence and the color confidence acquired in Steps 1)-3) on the basis of an SVM classifier; 5) judging whether moving speeds of the detection frames corresponding to motorcycles are within the threshold range or not, if the moving speeds of the detection frames corresponding to the motorcycles are within the threshold range, determining that detection results are correct, otherwise, determining that misinformation occurs. Compared with the prior art, the method has the advantages of low misinformation probability, simplicity in implementation and the like.
Description
Technical field
The present invention relates to traffic video detection field, more particularly, to a kind of motorcycle detection method of traffic video.
Background technology
Every year because motorcycle causes vehicle accident dead very many, motorcycle is because its speed is fast, poor performance, protective measure
It is weak, easily there is vehicle accident, and the injures and deaths after accident occurs are extremely serious, cause mostly head injurieies, this is motorcycle thing
Therefore the high main cause of mortality rate, disability rate.Motorcycle, electric motor car, bicycle is captured in traffic video to traffic violations
Detection has many helps, for example, can decide whether traveling etc. on car lane.Conventional motorcycle detection algorithm is mainly adopted
The mode combined with SVM by HOG.The major advantage of the method is that speed is fast, but performance is unsatisfactory for demand in practical application, such as
Situations such as there is more wrong report, Some vehicles and lose.
The content of the invention
The purpose of the present invention be exactly in order to overcome defect that above-mentioned prior art is present and to provide a kind of misinformation probability low, real
Apply the motorcycle detection method of simple traffic video.
The purpose of the present invention can be achieved through the following technical solutions：
A kind of motorcycle detection method of traffic video, the method obtain multiple detections by the way of scanning based on window
Frame, specifically includes following steps：
1) HoG features are extracted to each detection block, obtains HoG confidence levels；
2) LBP features are extracted to each detection block, obtains LBP confidence levels；
3) color characteristic is extracted to each detection block, obtains color confidence level；
4) SVM classifier is based on, according to step 1) -3) the HoG confidence levels that obtain, LBP confidence levels, color confidence level obtain
Motorcycle testing result；
5) judge that the translational speed of the detection block corresponding to motorcycle then judges detection whether in threshold range, if so,
As a result it is correct, if it is not, being then judged to wrong report.
The step 1) and step 2) in, HoG features, LBP features are classified using SVM classifier respectively, and then
Respectively obtain HoG confidence levels and LBP confidence levels.
The step 3) it is specially：
301) rectangle frame of the detection inframe any two position for a detection block, is taken respectively, calculates two rectangles
The similarity of the color histogram of frame, and preserve；
302) repeat step 301) D time, obtains D group color feature vectors；
303) repeat step is finished until the color feature vector of all detection blocks is extracted 301) with 302)；
304) classified using AdaBoost algorithms, obtained color confidence level.
The step 302) in, the value of number of times D is 100K～1M.
The step 304) in, by AdaBoost graders in multiple Weak Classifiers provide whether detection block is pedestrian's
Judged result, is weighted averagely to the judged result of multiple Weak Classifiers, obtains final AdaBoost algorithm classification results.
The color histogram adopt hsv color space, wherein each Color Channel be divided into K it is interval.
The value of the K is 6.
The step 5) in, threshold range is obtained by the labeled data at crossing.
The method also includes：
Road surface grader is obtained using texton boost Algorithm for Training, detection block of the testing result for motorcycle is input into
In the road surface grader, in judging the detection block, whether motorcycle lower zone is road surface, and if so, then testing result is correct,
If it is not, then testing result mistake.
Compared with prior art, the present invention has advantages below：
(1) the HoG confidence levels of each detection block, LBP confidence levels, color confidence level are considered by the present invention, are rubbed
Motorcycle testing result high precision；
(2) present invention is combined the translational speed of motorcycle when detecting, reduces wrong report；
(3) present invention also further improves accuracy of detection by road surface grader, effectively reduces misinformation probability；
(4) the present invention program is simple, it is easy to implement；
(5) the inventive method is applicable to the detection of motorcycle, electric motor car, bicycle, applied widely.
Description of the drawings
Fig. 1 is the schematic flow sheet of the present invention.
Specific embodiment
The present invention is described in detail with specific embodiment below in conjunction with the accompanying drawings.The present embodiment is with technical solution of the present invention
Premised on implemented, give detailed embodiment and specific operating process, but protection scope of the present invention be not limited to
Following embodiments.
Embodiment 1
As shown in figure 1, the present embodiment also provides a kind of motorcycle detection method of traffic video, the method is using based on window
The mode of mouth scanning obtains multiple detection blocks, can be applied to motorcycle, electric motor car, the detection of bicycle simultaneously.The method is concrete
Comprise the following steps：
Step S1, is extracted HoG features to each detection block, is classified using SVM classifier, obtains HoG confidence levels.
Step S2, is extracted LBP features to each detection block, is classified using SVM classifier, obtains LBP confidence levels.
Step S1 and step S2 are prior art conventional means.
Step S3, extracts color characteristic to each detection block, obtains color confidence level.The car sound color of motorcycle and motor
The clothes color of driver pedestrian has similarity：There can be a variety of colors, but in most cases, the color of itself is similar
's.According to the information, it is proposed that the feature of color self-similarity.Specially：
301) for a detection block, such as size is 32*64, takes the rectangle frame of the detection inframe any two position, such as
Size is 8*8, calculates the similarity of the color histogram of two rectangle frames, and preserves；
The color histogram adopts hsv color space, wherein each Color Channel be divided into K it is interval, the value of K is 6.
The computing formula of the similarity of the color histogram of two rectangle frames is：
piAnd qjThe rectangle frame of two positions, h is represented respectivelyi(k) and hjK () is p respectivelyiAnd qjColor histogram, k is
Histogram number, each color road be divided into 6 it is interval, a total of 216 are interval.
302) repeat step 301) D time, obtains D group color feature vectors.Optimum is selected as AdaBoost has
The characteristics of feature, therefore the value of D is to be the bigger the better, and can so cover various situations.For the consideration of training speed,
The span of final D is between 100K～1M.
303) repeat step is finished until the color feature vector of all detection blocks is extracted 301) with 302).
304) classified using AdaBoost algorithms, obtained color confidence level.It is multiple in by AdaBoost graders
Weak Classifier provides the judged result whether detection block is motorcycle, the judged result of multiple Weak Classifiers is weighted flat
, final AdaBoost algorithm classification results are obtained.
Step S4, based on SVM classifier, the HoG confidence levels obtained according to step S1-S3, LBP confidence levels, color confidence
Degree obtains motorcycle testing result.
Step S5, judges that the translational speed of detection block corresponding to motorcycle, whether in threshold range, if so, then judges
Testing result is correct, if it is not, being then judged to wrong report.Threshold range is obtained by the labeled data at crossing, i.e., motorcycle is on picture
Translational speed scope.
Embodiment 2
With reference to shown in Fig. 1, in order to further improve accuracy of detection, the detection method that the present embodiment is provided also includes last
One step：
Road surface grader is obtained using texture and color characteristic training, the present embodiment adopts texton boost algorithms, calculates
Method finally understand output image in be road surface region；Detection block of the testing result for motorcycle is input into into the road surface grader
In, in judging the detection block, whether motorcycle lower zone is road surface, and if so, then testing result is correct, if it is not, then testing result
Mistake.
Remaining is with embodiment 1.
Claims (8)
1. the motorcycle detection method of a kind of traffic video, it is characterised in that the method is obtained by the way of being scanned based on window
Multiple detection blocks are taken, following steps are specifically included：
1) HoG features are extracted to each detection block, obtains HoG confidence levels；
2) LBP features are extracted to each detection block, obtains LBP confidence levels；
3) color characteristic is extracted to each detection block, obtains color confidence level；
4) SVM classifier is based on, according to step 1) -3) the HoG confidence levels that obtain, LBP confidence levels, color confidence level obtain motor
Car testing result；
5) judge that the translational speed of the detection block corresponding to motorcycle then judges testing result whether in threshold range, if so,
Correctly, if it is not, being then judged to wrong report；
The step 3) it is specially：
301) rectangle frame of the detection inframe any two position for a detection block, is taken respectively, calculates two rectangle frames
The similarity of color histogram, and preserve；
302) repeat step 301) D time, obtains D group color feature vectors；
303) repeat step is finished until the color feature vector of all detection blocks is extracted 301) with 302)；
304) classified using AdaBoost algorithms, obtained color confidence level.
2. the motorcycle detection method of traffic video according to claim 1, it is characterised in that the step 1) and step
2) in, HoG features, LBP features are classified using SVM classifier respectively, and then respectively obtains HoG confidence levels and LBP puts
Reliability.
3. the motorcycle detection method of traffic video according to claim 1, it is characterised in that the step 302) in,
The value of number of times D is 105～106。
4. the motorcycle detection method of traffic video according to claim 1, it is characterised in that the step 304) in,
Multiple Weak Classifiers in by AdaBoost graders provide the judged result whether detection block is pedestrian, to multiple Weak Classifiers
Judged result be weighted averagely, obtain final AdaBoost algorithm classification results.
5. the motorcycle detection method of traffic video according to claim 1, it is characterised in that the color histogram is adopted
Use hsv color space, wherein each Color Channel be divided into K it is interval.
6. the motorcycle detection method of traffic video according to claim 5, it is characterised in that the value of the K is 6.
7. the motorcycle detection method of traffic video according to claim 1, it is characterised in that the step 5) in, threshold
Value scope is obtained by the labeled data at crossing.
8. the motorcycle detection method of traffic video according to claim 1, it is characterised in that the method also includes：
Road surface grader is obtained using texton boost Algorithm for Training, testing result is described for the detection block input of motorcycle
In the grader of road surface, in judging the detection block, whether motorcycle lower zone is road surface, and if so, then testing result is correct, if it is not,
Then testing result mistake.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510102547.7A CN104658254B (en) | 2015-03-09 | 2015-03-09 | Motorcycle detection method for traffic videos |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510102547.7A CN104658254B (en) | 2015-03-09 | 2015-03-09 | Motorcycle detection method for traffic videos |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104658254A CN104658254A (en) | 2015-05-27 |
| CN104658254B true CN104658254B (en) | 2017-03-22 |
Family
ID=53249316
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510102547.7A Active CN104658254B (en) | 2015-03-09 | 2015-03-09 | Motorcycle detection method for traffic videos |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104658254B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017107188A1 (en) * | 2015-12-25 | 2017-06-29 | 中国科学院深圳先进技术研究院 | Method and apparatus for rapidly recognizing video classification |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2002008189A (en) * | 2000-06-22 | 2002-01-11 | Matsushita Electric Ind Co Ltd | Vehicle detector and vehicle detection method |
| CN101601088B (en) * | 2007-09-11 | 2012-05-30 | 松下电器产业株式会社 | Sound judging device, sound sensing device, and sound judging method |
| US8681222B2 (en) * | 2010-12-08 | 2014-03-25 | GM Global Technology Operations LLC | Adaptation for clear path detection with additional classifiers |
| CN104200668B (en) * | 2014-07-28 | 2017-07-11 | 四川大学 | A kind of motorcycle based on graphical analysis is not helmeted violation event detection method |
| CN104217217B (en) * | 2014-09-02 | 2017-06-30 | 武汉睿智视讯科技有限公司 | A kind of vehicle mark object detecting method and system based on two layers of classified |
-
2015
- 2015-03-09 CN CN201510102547.7A patent/CN104658254B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN104658254A (en) | 2015-05-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2019196130A1 (en) | Classifier training method and device for vehicle-mounted thermal imaging pedestrian detection | |
| US10025998B1 (en) | Object detection using candidate object alignment | |
| Xu et al. | Detection of sudden pedestrian crossings for driving assistance systems | |
| WO2020173022A1 (en) | Vehicle violation identifying method, server and storage medium | |
| US8902053B2 (en) | Method and system for lane departure warning | |
| CN108549864B (en) | Vehicle-mounted thermal imaging pedestrian detection-oriented region-of-interest filtering method and device | |
| US9904856B2 (en) | Method and apparatus for detecting target object in blind area of vehicle | |
| Huang et al. | Vehicle detection and inter-vehicle distance estimation using single-lens video camera on urban/suburb roads | |
| CN102509098B (en) | Fisheye image vehicle identification method | |
| US20170032676A1 (en) | System for detecting pedestrians by fusing color and depth information | |
| CN103530600B (en) | Licence plate recognition method under complex illumination and system | |
| CN106778736B (en) | Robust license plate recognition method and system | |
| CN104657724A (en) | Method for detecting pedestrians in traffic videos | |
| CN103530640B (en) | Unlicensed vehicle checking method based on AdaBoost Yu SVM | |
| CN103268489A (en) | Motor vehicle plate identification method based on sliding window searching | |
| CN107578048B (en) | Vehicle type rough classification-based far-view scene vehicle detection method | |
| CN104200207A (en) | License plate recognition method based on Hidden Markov models | |
| CN106919939B (en) | A kind of traffic signboard tracks and identifies method and system | |
| CN105469124A (en) | Traffic sign classification method | |
| CN103106409A (en) | Composite character extraction method aiming at head shoulder detection | |
| Rabiu | Vehicle detection and classification for cluttered urban intersection | |
| CN106022242B (en) | Driver's making and receiving calls recognition methods in intelligent transportation system | |
| CN105426816A (en) | Method and device of processing face images | |
| Ingole et al. | Characters feature based Indian vehicle license plate detection and recognition | |
| KR102372296B1 (en) | Apparatus and method for recognize driving lane on image |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |